Anthony Scriffignano, senior vice president and chief data scientist at Dun & Bradstreet Inc., said CIOs should...
start to "pay attention to new nouns and verbs" in 2018.
He points to language development as evidence that AI will be an important topic for CIOs this year. New nouns, verbs and adjectives are beginning to emerge as companies talk about AI. Case in point, according to Scriffignano: cognitive computing, or AI agents that assist and advise humans. When CIOs encounter new language like this, they shouldn't brush it aside but instead ask, "What nuance does that word introduce and what does it really mean?" he said.
Thinking linguistically is just one of a list of AI topics Scriffignano said CIOs should keep an eye on in 2018. Here are eight additional AI topics he believes will make headlines in the new year.
1. Don't wait until it's too late to deal with shadow AI. The proliferation of cloud-based AI tools and services -- available to organizations at the swipe of a credit card -- means CIOs should expect AI projects to pop up throughout the enterprise, a phenomenon Scriffignano referred to as the federation of AI. He urged CIOs to build a strategy for handling AI projects not led by IT.
2. Regulations will have an impact on your AI projects. Cybersecurity, data sovereignty and data transfer regulations "influence the space of AI," Scriffignano said. Today, the General Data Protection Regulation (GDPR), which goes into effect in May, is top of mind for many organizations, and it includes a "right to explanation" mandate, a sticking point for machine learning models that operate inside a so-called black box. But soon enough, GDPR will be replaced by another regulation -- and by new forms of malfeasance. "When there's new regulation being contemplated, there are the people trying to comply with it and the people trying to figure out how to get around it," he said.
3. Cybercriminals will use AI to sharpen cyberattacks. Scriffignano encouraged CIOs to be on the lookout for "the amalgamation of AI and cyber anything." Take botnet attacks, for instance. "They're pretty stupid right now," he said, and require going out and searching for connected devices with static IP addresses. "I'm not the cyber guy, but I'm pretty sure that if these botnet attacks started to use flocking and swarming algorithms and started learning from their mistakes and redirecting their efforts based on how they're failing, they could succeed a lot quicker," he said. "That's terrifying."
4. The IoT-AI relationship is a double-edged sword. The internet of things (IoT) is largely a network of things connected to the internet. "These things don't have a very good ability to discover each other autonomously and have a conversation about what each of them does and how they might help each other," Scriffignano said. But thanks to AI, the ability for devices to communicate with and even learn from each other is coming, which will present new ways to delight customers but also new security and data privacy obstacles for CIOs. "We've got to be smarter about how these things connect to each other," he said, "and how the unintended consequences of them connecting to each other without us controlling those connections might cause mysterious things to happen."
5. Cognitive computing is not your run-of-the-mill AI. CIOs should think of cognitive computing as a new field of AI. "The idea behind cognitive computing is an AI agent who works alongside human experts and advises the expert and watches the degree to which the advice is taken or not taken," he said, "and modifies its advice accordingly." But today, cognitive is often used as a synonym for AI or slapped on to processes -- like your autocorrect function -- that don't rise to the level of cognitive computing. Scriffignano urges CIOs to pay attention to how cognitive develops in 2018 and not use it as catch-all term for AI.
6. Collaborate without commoditizing. Most CIOs won't be building an end-to-end AI solution, Scriffignano said. Instead, they'll use platforms and services that can provide functionality to the business, creating a new kind of partnership between organizations. Adding AI to the mix will "challenge the ways in which we collaborate because it will form new connections that are sort of accidental," he said. "If everything is formed as a service and ubiquitous and discoverable, then there's a tendency for things to get commoditized. So, we have to figure out how to collaborate without commoditizing."
7. Talent will continue to be a challenge. New skills are required to operate today's data-driven companies, a need that's exacerbated by AI. That doesn't mean CIOs can stop hiring for the skills they used to hire for -- data curators, analysts, modelers statisticians and methodologists are still needed in the modern enterprise. "But, increasingly, we need governance experts and problem formulators and detectives and visionaries and storytellers and diplomats," Scriffignano said. "So, the stakes have gone way up."
8. Watch out for autonomous AI. Autonomy means an AI agent is disconnected from a human or human-made mechanism that is telling it what to do. "Every AI agent has a goal, something it's trying to achieve," Scriffignano said. "If it's autonomous and the environment changes, it needs to have the ability to modify its goal" to be successful. But AI autonomy could lead companies into dangerous territory.